Research demo
AgentFlayer: a malicious Jira ticket makes Cursor exfiltrate repo secrets and AWS credentials
AIR-2025-0072 AIR record · 2025-08-cursor-agentflayer-jira-ticket
From the supporting source
Using terms like 'apples' instead of 'API keys,' was all that was required to bypass the model's built-in controls.labs.zenity.io
The record
Disclosed
2025-08-01
Vendor
Anysphere (Cursor); Atlassian (Jira)
Product
Cursor IDE with Jira MCP integration
Risk domain
Privacy › Data exfiltration channels
Also touches
Credential theft, Confused deputy, Supply chain vulnerabilities
System type
Generative model
Who acted
Agent acted — Cursor read the ticket, read the secrets and sent them out
Attack surface
Coding agent
Autonomy level
Semi-autonomous
Initial vector
Indirect prompt injection mcp tool output
Impact type
Data exfiltration, Credential theft
Confidence
High
Submitted by
EAI Internal
Crosswalk
- OWASP Agentic Top 10 (2026)
- ASI02 – Tool Misuse and Exploitation
- MITRE ATLAS
- AML.T0024 – Exfiltration via AI Inference API
- NIST AI RMF
- MAP 4.2
- EU AI Act
- Article 10
- ISO/IEC
- TR 24028; 23894
- AIUC-1
- A. Data & Privacy
- Scope
- tool misuse
- Mitigation
- Monitor network traffic for anomalous patterns; Restrict egress destinations.
- Taxonomy
- Enkrypt AI agent risk taxonomy · Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery
Blast radius
Cursor users with a Jira MCP server connected to an externally-writable issue tracker
Capability abused
Defense & remediation
Recorded defense
Cursor model-level content controls
Guardrail kinds
Unknown
Guardrail outcome
Silent
Fix shipped
none; Cursor characterised untrusted-data MCP servers as a known risk
Taxonomy & references
Curator notes
Marina Simakov, Zenity Labs, 1 Aug 2025. Cursor's stated position: 'MCP servers, especially ones that connect to untrusted data sources, present a serious risk to users.' Distinct from Check Point's MCPoison (CVE-2025-54136) and Aim Security's CurXecute (CVE-2025-54135), which are separate Cursor findings belonging with the responsible-disclosure records.
Supporting sources
Cite as
Enkrypt AI. "AgentFlayer: a malicious Jira ticket makes Cursor exfiltrate repo secrets and AWS credentials". AIR, AIR-2025-0072. https://agentincident.vercel.app/i/AIR-2025-0072/ (accessed 2026-10-19).